List your company

Instruction on IFEval (avg@32 accuracy)

46.2Avg@32 AccuracyIBTPO
20.92827.48934.0540.611May 27, 2026

Evaluation results

MethodLinks
IBTPOBase Model=Qwen3-8B-Base2026.0546.2
GRPO w/ Clip-higherBase Model=Qwen3-8B-Base2026.0545.4
IBROBase Model=Qwen3-8B-Base2026.0544.7
GRPO w/ Entropy Reg†Base Model=Qwen3-8B-Base2026.0544.2
Initial ModelBase Model=Qwen3-8B-Base2026.0543.8
TreeRLBase Model=Qwen3-8B-Base2026.0542.5
TreePOBase Model=Qwen3-8B-Base2026.0542.2
Vanilla GRPOBase Model=Qwen3-8B-Base2026.0542
IBTPOBase Model=Qwen3-1.7B-Base2026.0526.9
IBROBase Model=Qwen3-1.7B-Base2026.0525.4
Vanilla GRPOBase Model=Qwen3-1.7B-Base2026.0524
GRPO w/ Clip-higherBase Model=Qwen3-1.7B-Base2026.0523.7
GRPO w/ Entropy Reg†Base Model=Qwen3-1.7B-Base2026.0523.7
TreeRLBase Model=Qwen3-1.7B-Base2026.0523.5
TreePOBase Model=Qwen3-1.7B-Base2026.0523.2
Initial ModelBase Model=Qwen3-1.7B-Base2026.0521.9